Empirical performance of rough volatility models

Abstract

In this thesis, we examine the empirical performance of rough volatility models within the S&P 500 and KOSPI 200 Index options markets. We thoroughly investigate the rough Heston and rough Bergomi models, employing a neural network-based calibration method to expedite calibration times. Our findings suggest that rough volatility models generally outperform the classical Heston model in both in-sample and out-of-sample pricing performance. Their hedging performance, however, does not consistently exhibit improvement, except for a notable enhancement in KOSPI 200 put options. This research contributes to filling the void in empirical studies on rough volatility models, robustly validating their superior performance in two prominent options markets.